LED Bulbs

[TjW]

Fluorescents do come in different colors. Here is a warm white vs cool white in my kitchen.

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_1178252f27a66e402d.png[/qimg]

Electronic ballast that doubles the AC frequency will also get rid of the (perceptible) flickering, and any buzzing noise.

I recently changed out my T12 (1.5 inch diameter) fluorescent tubes for T8 (1 inch diameter). The ballast on the T12s had gone. The T8s seem to give the same amount of light from 32W bulbs that the T12s gave from 40W.
(not measured power or lumens, just nominal values and my perception.)

The T8 electronic ballasts that I bought switch at ~40kHz. I don't think anyone's going to be seeing that flicker, or hear that buzz.

 

[Brian-M]

Yesterday , a second bulb failed in the same ceiling fitting.
This seems unlikely to be coincidence. If there is a reason LEDs should not be used in multi-bulb fittings, I'd think we need to know about it. These bulbs are rated for 25000 hours Thats two falures at about 4% of the claimed life. I'm not impressed.

Just guess here... we know that LEDs can fail if they get too hot, and overheating is a major problem. If the multi-bulb ceiling fitting is enclosed, maybe overheating is the problem?

ETA:

The most serious effect of heating LEDs is a reduction of lifetime. When overheating damages an area of a semiconductor crystal, it is likely to offer more electrical resistance. Then the area of increased electrical resistance is even more likely to overheat again, causing additional damage. Thermal degradation can reduce the operational lifetime from a potential of tens of thousands of hours to just a few thousand or even less.

http://www.ehow.com/info_8303658_effects-heat-led-lights.html

 

[Soapy Sam]

It's more likely that the problem is with your light fitting, not the bulbs. LEDs are a lot more sensitive to power supply problems than incandescents, since they have more than just a resistor in them, so just because you didn't notice a problem before doesn't mean there wasn't one.

I don't see any cautions about " sensitive to power problems" on the box. Maybe when they sell something 20 times the price of the alternative model, there should be.
The sparky who is going to fit the new lights in the conservatory was here today. He was surprised by the failures, but found nothing wrong with the power supply to the fitting.
The light fitting was in the house when we arrived in April 2011. It had 6 halogen GU10s.
Back in April, I wanted to replace 5 12V MR16s in the bathroom with 240v LEDs. Despite manufacturer's assurances that the recessed fittings were compatible with all types of GU10 bulb, it turned out (after I had enlarged holes in the ceiling to take the fittings) that the bulbs did not fit because they had cooling radiator ribs which hung up as I tried to install the bulbs.

So I moved five of the halogens from the bedroom fitting to the bathroom and used the new LEDs in the bedroom. This is actually my computer room, so the lights get more use than the bathroom ones anyway. The halogens worked fine in the new fittings and 10 months later, they still are working fine, as is the single one still in the bedroom. If this was a heat issue, I'd expect the halogen to fail first as it is far hotter than the LEDs.

As you know, GU10s are spotlights, designed for narrow, metal fittings which do get hot. I find the halogen fitting too hot to hold for more than half a minute, but I could hold the LED fittings indefinitely. It can't be heat. Given the halogens never had any problems in that fitting, I really can't think of any other reason it would be faulty. There's no discernible vibration
It's odd.
I like the light from the LEDs. I just can't afford to replace them at this rate.

ETA- Of course it can be heat and probably is, because there's no other reason, and the fact the ES50s have cooling fins is itself suspicious- but by GU 10 standards this is actually a well ventilated fitting. If these bulbs overheat at far lower temperatures than halogens, then I truly think that's something that should be made very clear. They are sold as suitable for existing light fittings. If , in fact , they are not suitable, then the cost per time calculation must include the replacement of fittings and any associated redecoration. That could greatly multiply the time required to get your money back in power savings.
Here's a pic of the fitting, deliberately underexposed to show colour.
Left end, 1 35W halogen. Middle three are 7.5w ES50s. On the right, 2 cool white 4W LEDs (replacing the failures).Bulbs are almost 1 foot below the ceiling. The stainless steel holders are vented and only the halogen is hot to the touch.

 

[Soapy Sam]

I haven't tried warm white yet, but I recently replaced neutral white fluorescents with neutral white LEDs and the color is pretty much the same.

Cool white LEDs are the most efficient. The warmer you go, the less efficiency you get.

The use of "cool" in this context confused me. My wife (an artist) thinks of blue light as "cold". I assume blue light is always going to be a higher temperature than red light. I expected them to be yellowish.

I just stuck a thermometer on the three bulb types in the pic above. It's a room thermo, not contact, so I don't claim accuracy, but it gives a relative idea.

The halogen stabilised at 41C after 1 minute. The ES50 warm white LED 26C, the cool white 25. Ambient air temperature is 20 C (68F). The appearance of the two cool whites is distinctly bluish.
One big difference is that the ES50 has only 3 actual diodes, while the cool whites have 19 each, so the cool whites are actually putting out much less heat each than the warm whites.

Surely it is the electronics rather than the LEDs that are failing. I assume they are multi-LED bulbs, so if LEDs were failing a whole bulb would not fail at once. I suspect that in specifying lifetime they are only considering LED life and pairing that with cheap DC conversion that is not well tested.

Yes, what happens is one diode fails (as you say , probably due to the DC supply electronics failing) that could be lived with- but the remaining two diodes start pulsing like a strobe unit. Quite intolerable.

So far none of the 6 cool whites I fitted (4 in a kitchen light and the two in the pic above) have given problems. I don't know if in their case , the remaining 18 diodes would work normally.

I prefer the warm white light, which is slightly pinker than the halogen, but not much. As you can see, the cool whites are distinctly different.

 

[Modified]

I recently changed out my T12 (1.5 inch diameter) fluorescent tubes for T8 (1 inch diameter). The ballast on the T12s had gone. The T8s seem to give the same amount of light from 32W bulbs that the T12s gave from 40W.
(not measured power or lumens, just nominal values and my perception.)

That sounds about right. Ballast failing is the reason I went from T12s to LEDs. I'm getting about twice the efficiency, but T8 to LED would get you only a small improvement.

 

[bruto]

I just did a very quick and dirty comparison between bulbs here. I set my digital camera's white balance to "incandescent," and shot a white-ish object under a 50 watt incandescent flood lamp, a standard sort of CFL in a lamp, and a 9-watt (supposedly 60 equivalent) LED in a ceiling fixture. Relative brightness is not shown since the camera's exposure is automatic. As you can see, the "cool white" LED comes tolerably close to the incandescent, with the CFL considerably warmer.

Of course, in real life, the incandescent light is much yellower than natural light, but assuming that it is the preferred indoor temperature, this gives some idea of how the bulbs compare.

 

[Brian-M]

a standard sort of CFL in a lamp,

What do you mean by "standard"? The CFLs you buy in the store can be warm white, cool white or daylight. You have to check the box before you buy them to make sure it's the kind you want.

The use of "cool" in this context confused me. My wife (an artist) thinks of blue light as "cold". I assume blue light is always going to be a higher temperature than red light. I expected them to be yellowish.

You're both right. "Cool" white has a higher color temperature than warm white and is also bluer than warm white. But the color temperature is only related to actual temperature in incandescent globes. In fluros and LEDs it's a product of the mixture of phosphorescent substances used to produce the light.

 

[BenBurch]

Still really happy with my Switch Lighting bulbs. Have three of theirs now. One commercial line 3-way bulb and two "60w" equivalent consumer-grade bulbs. Actually brighter than their incandescent equivalents, but I understand that LEDs degrade a few percent over their lifetimes.

 

[bruto]

What do you mean by "standard"? The CFLs you buy in the store can be warm white, cool white or daylight. You have to check the box before you buy them to make sure it's the kind you want.

I'm not sure. There's no designation on the bulb, and I don't remember where I got it. Here in Vermont, where the power company is pushing for conservation, CFL bulbs are dirt cheap and often subsidized, rarely more than a buck apiece, and sometimes free or nearly so. I don't keep much track of them. Some last forever, some fail quickly, some turn up in used fixtures. I have a big pot of them, and draw on it when I need a bulb, but take little care about more than the approximate wattage. I'm guessing warm white or something near to it, since it's yellower than the incandescent, which I'm guessing is cool white. The wattage of the cfl bulb in question was 26, I think, and it's supposed to be equivalent to about a 75.

 

[Prometheus]

I have one LED bulb that's burned about 12 hours a day for 4 and half years now.

 

[Iamme]

If anyone has trouble with premature bulb failure, carefully analyze the lamp socket. Look for excessive flattened/pitted center tang (hot wire connects to this), and carefully look at the socket base rivots(neutral wire connects to this) for tan/black flash marks. You should not see any of that. If you do, the socket must be replaced.
.
Regarding 26wcfl = 75winc, i dont think so since 13wcfl = 60winc, and 23wcfl = 100winc equivalent. So my guess is 75w incandescent is more like about 18watts cfl.

 

[bruto]

If anyone has trouble with premature bulb failure, carefully analyze the lamp socket. Look for excessive flattened/pitted center tang (hot wire connects to this), and carefully look at the socket base rivots(neutral wire connects to this) for tan/black flash marks. You should not see any of that. If you do, the socket must be replaced.
.
Regarding 26wcfl = 75winc, i dont think so since 13wcfl = 60winc, and 23wcfl = 100winc equivalent. So my guess is 75w incandescent is more like about 18watts cfl.

My supposedly hundred watt equivalent CFL's seem to be 26, but these things vary. Two different brands of LED 60's I have are 9 and 11.

You can pull up a flattened center tang at least a few times before it gets too lazy to stay up, and it will help. Sockets with a bad center contact will overheat the bulb. This is also true, by the way, for screw-in fuses, though they do not have sprung center tangs. If the center contact gets dirty and pitted, the fuse will overheat and blow frequently. Lamp sockets usually have a spring so you don't have to screw the bulb in so tight and risk breakage, and to allow for some thermal changes.

 

[a_unique_person]

25 watts cfl is as good as a 100 incandescent for me. The new technology ones are also faster to fire up.

 

[Brian-M]

If anyone has trouble with premature bulb failure, carefully analyze the lamp socket. Look for excessive flattened/pitted center tang (hot wire connects to this), and carefully look at the socket base rivots(neutral wire connects to this) for tan/black flash marks. You should not see any of that. If you do, the socket must be replaced.

Of course, that description is specific to bulbs that fit into Edison-Screw sockets. But now you have me wondering if the bayonet-cap sockets that are widely used over here are more or less prone to hot connections.

I don't know how common they are over there, but every place I've lived in here uses Bayonet-Cap fittings. The metal part of the base of the bulb that "screws" in to the socket isn't connected to power, you have two contacts at the base of the... stem?... that press onto spring-loaded pins. They're easier to insert and remove than the screw kind. Just press the bulb further into the socket and turn your wrist about a quarter-turn anti-clockwise and it pops right out.

I wonder if there have ever been studies done comparing the reliability of these two systems.

 

[bruto]

Of course, that description is specific to bulbs that fit into Edison-Screw sockets. But now you have me wondering if the bayonet-cap sockets that are widely used over here are more or less prone to hot connections.

I don't know how common they are over there, but every place I've lived in here uses Bayonet-Cap fittings. The metal part of the base of the bulb that "screws" in to the socket isn't connected to power, you have two contacts at the base of the... stem?... that press onto spring-loaded pins. They're easier to insert and remove than the screw kind. Just press the bulb further into the socket and turn your wrist about a quarter-turn anti-clockwise and it pops right out.

I wonder if there have ever been studies done comparing the reliability of these two systems.

In the US we rarely if ever see bayonet fittings for line voltage bulbs. You occasionally see them in specialized applications such as vacuum cleaner headlights, sewing machines and the like, but most bayonet bulbs are low voltage. Of course they're common in cars and the like, and when one does see this kind of fitting, it's usually similar, though most use a single center conductor and use the base as the ground conductor. The center contacts on these are usually a spring-loaded disk with a conductor in the middle, and less likely to suffer from the problem found with Edison-base bulbs. In the commonest version of the Edison base, the center conductor is simply a springy piece of some alloy, which eventually loses its spring. In addition, of course, a bayonet bulb is always inserted correctly if it's inserted at all, where a screw base can be put in loose.

As to reliability, you won't see screw-in bulbs in automotive applications, probably because they would vibrate loose. Bayonets have recently begun to be superseded by push-in bases, presumably because they're cheaper and perhaps a bit less susceptible to corrosion.

 

[Psi Baba]

Y'know, this is all starting to sound like the thread title should be changed to "What if Microsoft invented light bulbs?" It shouldn't be this complicated.

 

[Soapy Sam]

I opened one of the two Sylvania ES50s that failed.
The bulbs have a well made, quite heavy heat sink, but when I removed the plastic lens cover, I found the connection of the emitters to the heat sink is questionable.
The three LEDs are surface mounted on a triangular plastic circuit board. The two power prongs slot through holes cut in the board, into metal clamp type connectors.

The board connects to the heat sink only via its surface which has been very patchily coated with what I assume was a thermally conductive contact paste.
As you can see from the photo below, the application of that is far from "generous" or even uniform. If this was a CPU, I'd expect it to die young.

There are a number of interesting Youtube videos in which different LED bulbs are taken apart. It's clear quality varies hugely, but this is a pricy bulb, made by a major manufacturer and I don't feel that 40% failure at 4% of the advertised life is remotely acceptable. Next one to fail will be replaced by a halogen.

http://www.internationalskeptics.com/forums/picture.php?albumid=425&pictureid=8626

 

[Dr. Keith]

It's clear quality varies hugely, but this is a pricy bulb, made by a major manufacturer and I don't feel that 40% failure at 4% of the advertised life is remotely acceptable. Next one to fail will be replaced by a halogen.

Have you contacted Sylvania to see if they will replace them?

 

[Soapy Sam]

Have you contacted Sylvania to see if they will replace them?

I emailed them when number 1 failed.

4 days later, number 2 failed.
I emailed them again.

2 days after that, I received a single bulb in an envelope, no note, no comment, nothing. So it may have been in response to failure 1 or 2.

I emailed again, giving details of the light fitting and asking for comment. No reply.

Tomorrow I plan to phone them and try to talk to someone technical. I will offer to send the photo above, or the remaining failed bulb (I already did offer). While the packaging displays the 25000 hour figure (and also says 25 YEARS) it does not mention a warranty. I'd say something that lasts 4% of it's planned life should be under warranty, but I doubt light bulbs historically carried much of a warranty.

If the realistic likely life is 1000 hours, I don't actually want to replace them. I want to find something better. I found an interesting Youtube video that seems to suggest the 3-LED format has not been living up to expected lifetimes and is being replaced by bulbs with a larger number of lower power LEDs. The 4W LumiLife I replaced the first ES50 with has 19 LEDs. Of course, I don't know yet how long those will last...

 

[BenBurch]

Still having zero problems with my lamps from Switch Lighting.